RESEARCH Open Access

Proteomic patterns analysis with multivariatecalculations as a promising tool for promptdifferentiation of early stage lung tissue withcancer and unchanged tissue materialPiotr Waloszczyk1, Tomasz Janus1*, Jacek Alchimowicz2, Tomasz Grodzki2, Krzysztof Borowiak1

Abstract

Background: Lung cancer diagnosis in tissue material with commonly used histological techniques is sometimesinconvenient and in a number of cases leads to ambiguous conclusions. Frequently advanced immunostainingtechniques have to be employed, yet they are both time consuming and limited. In this study a proteomicapproach is presented which may help provide unambiguous pathologic diagnosis of tissue material.

Methods: Lung tissue material found to be pathologically changed was prepared to isolate proteome with fastand non selective procedure. Isolated peptides and proteins in ranging from 3.5 to 20 kDa were analysed directlyusing high resolution mass spectrometer (MALDI-TOF/TOF) with sinapic acid as a matrix. Recorded complex spectraof a single run were then analyzed with multivariate statistical analysis algorithms (principle component analysis,classification methods). In the applied protocol we focused on obtaining the spectra richest in protein signalsconstituting a pattern of change within the sample containing detailed information about its protein composition.Advanced statistical methods were to indicate differences between examined groups.

Results: Obtained results indicate changes in proteome profiles of changed tissues in comparison tophysiologically unchanged material (control group) which were reflected in the result of principle componentanalysis (PCA). Points representing spectra of control group were located in different areas of multidimensionalspace and were less diffused in comparison to cancer tissues. Three different classification algorithms showedrecognition capability of 100% regarding classification of examined material into an appropriate group.

Conclusion: The application of the presented protocol and method enabled finding pathological changes in tissuematerial regardless of localization and size of abnormalities in the sample volume. Proteomic profile as a complex,rich in signals spectrum of proteins can be expressed as a single point in multidimensional space and thananalysed using advanced statistical methods. This approach seems to provide more precise information about apathology and may be considered in futer evaluation of biomarkers for clinical applications in different pathology.Multiparameter statistical methods may be helpful in elucidation of newly expressed sensitive biomarkers definedas many factors “in one point”.

* Correspondence: tjanus@sci.pam.szczecin.pl1Department of Toxicology and Molecular Pathobiochemistry, PomeranianMedical University, 70-204 Szczecin, PolandFull list of author information is available at the end of the article

Waloszczyk et al. Diagnostic Pathology 2011, 6:22http://www.diagnosticpathology.org/content/6/1/22

© 2011 Waloszczyk et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the CreativeCommons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, andreproduction in any medium, provided the original work is properly cited.

mailto:tjanus@sci.pam.szczecin.plhttp://creativecommons.org/licenses/by/2.0

BackgroundCancer diagnosis based on standard histological meth-ods is widely described and used in medicine. However,most of the procedures derive from a subjective assess-ment of observed changes and in some cases may beinconclusive. In practical terms, new, objective methodsin cancer diagnosis are still needed especially in respectof their sensitive. Moreover, tumour growth in its earlystages is restricted to only a part of the tissue and insome cases it may be overlooked. Recently, proteomicshas proved to be a valuable approach in biomarkerdetection and finding sensitive parameters which indi-cate disease process [1-4]. The combination of spectro-scopic methods of high resolution (mass spectroscopy,nuclear magnetic resonance) with advanced statisticalmethods leads to an increased likelihood of developingnew applications for diagnostic purposes [5-9]. Molecu-lar imaging techniques described in literature as well asprofiling and detailed characterisation of biomarkermethods tend to prove their usefulness [10-12]. How-ever, many protocols while focusing on details and theo-retical aspects, do not translate directly into practicalapplications since a large number of samples need to betested to validate them. Molecular imaging with massspectrometry may offer ample possibilities in practicalusage yet in the future rather than at present, becauseof extent of time necessary for analysis, workload ofinstruments, and poor image resolution in comparisonto other methods [13-15].Our work suggests an option of proteomic profiling of

biological materials which focuses on obtaining spectrarich in proteins that may reveal the proteomic composi-tion of tissue. Using multivariate statistical methodseach complex spectrum can be shown as a single pointin multidimensional space (as a single parameter), thenanalysed in respect of localization in this space. The sin-gle point in multidimensional space become a specificbiomarker if its location does not match the controlregion. Even a slight change in tissue structure and thenature of pathology of any origin should be reflected inproteomic profile, provided that profiles richest in func-tional proteome of mass range of 3.5 - 20 kDa (withouthigh abundant proteins - large proteins) are obtained.These profiles may correspond to a type of the patholo-gical change (cancer, inflammation) as well as to normalprofiles. The obtained profiles form something like a“proteomic fingerprint” and allow for classification oftested materials into appropriate groups using statisticalmethods. Tissue models have to be built for a knownlung tissue structure (normal, changed: cancer, inflam-mation) using supervised classification methods. Thesemethods indicate also signals typical for both normaland pathological tissue (table 1). Once appropriate

models are calculated, any unknown lung tissue materialcan be measured and classified into defined groups (pre-viously built models). The appearance of signals typicalfor a pathology results in classification of examined tis-sue into an appropriate model. This appears to be asimple protocol without complex, time consuming andexpensive procedures which seems to be relevant con-sidering a large number of samples needed for futuremodels validation.

MethodsMaterialsSegments of specimens collected for an intraoperativeconsultation were examined. Only lung cancer suspectedsamples were taken into consideration. Diagnosis wasmade basing on segments prepared with a standard crio-static technique followed by revision on the basis of par-affin method from specimens stained with hematoxilinand eozine (HE) and complemented with immunohisto-chemical method. Pathological changes in tissues wereclassified according to The Word Health Organizationclassification. Types of pathology are listed in table 2.All the material for proteomic examination was col-

lected from tissue fragments which did not show necro-sis in volume in a macroscopic examination.The control group (15 different cases) consisted of

pathologically unchanged fragments of the lung tissueobtained from lungs removed during surgery on traumacases. The material did not show any histologicalchanges in histological examination that followed.

MethodsMass spectra were recorded in a linear mode of 3.1 -20 kDa with a high resolution mass spectrometer oftime of flight type (MALDI-TOF/TOF) Bruker versionAutoflex III Smartbeam with 200 Hz laser technology.Sinapic acid (SA) of proteomic grade (Bruker) was usedas a matrix. 400 mg of lung tissue was directly frozen at-80°C. Next, just before measurement, the material wasdefrozen, homogenized with 0.7 ml 0.9% NaCl solutionand centrifuged at 13,000 rpm for 10 minutes. Theobtained supernatant of 0.6 ml was treated with 50μl of10% trifluoracetic acid (TFA) and centrifuged again andthen it was transferred to Amicon Ultracel (Millipore)centrifugal filter of 3 kDa selectivity and centifuged at13,000 rpm to decrease the volume to 50 μl. The con-centrated solution containing proteins and peptides wasdesalted by twice repeated centrifugation in the sameconditions with 450 μl of deionized water. Concentratedand desalted material of 10 μl was transferred to a coni-cal tube and gently mixed with 20 μl of saturated SA(SA in water with 0.1% TFA and acetonitryl with 0.1%TFA at a ratio of 60:40). The obtained mixture was

Waloszczyk et al. Diagnostic Pathology 2011, 6:22http://www.diagnosticpathology.org/content/6/1/22

Page 2 of 7

directly applied onto standard steel target and analyzedfollowing solvent evaporation.The instrument operated under control of FlexControl

software (Bruker).All the statistical calculations were done using Clin-

ProtTols ver. 2.2 software (Bruker). Proteomic profiles

were analyzed with principle component analysis algo-rithm (PCA) to observe differences in proteome profilesof particular groups. Supervised classification algorithms,Support Vector Machine (SVM), Genetic Algorithm(GA) and Supervised Neural Network (SNN) were usedfor classification of tissue signals into appropriate groups.

Table 1 Peak statistics - comparison of statistically important (p value of T-test/ANOVA < 0.05) peaks for dataseparation and recognition capability (biomarkers candidates)

Mass (Da) Mean (Control) Mean (Pathological) P value T-Test/ANOVA P value Wilcoxon/Kruskal-Wallis Difference Average

8616.14 0.32 1.45 < 0.000001 0.00000847 < 0.000001

6228.05 1.55 4.07 < 0.000001 0.00152 < 0.000001

5759.35 1.54 2.65 < 0.000001 0.00017 0.000773

3588.09 1.99 4.82 < 0.000001 0.00017 < 0.000001

7596.85 0.81 2.11 < 0.000001 0.00154 < 0.000001

5201.26 1.93 3.87 < 0.000001 0.000366 < 0.000001

4156.39 2.08 4.87 < 0.000001 0.00017 0.0000021

9554.61 0.78 1.99 0.00000228 0.11 < 0.000001

12350.75 0.25 1.22 0.00000228 0.0000289 < 0.000001

3603.83 2.11 4.11 0.0000024 0.0162 < 0.000001

6723.40 20.93 8.96 0.00000228 0.00000855 0.000101

11297.49 0.45 0.97 0.00000228 0.00807 < 0.000001

10268.09 0.59 1.56 0.00000341 0.000314 < 0.000001

10525.25 0.3 1.19 0.00000695 0.00017 < 0.000001

5800.39 1.79 3.82 0.00000122 0.195 < 0.000001

5910.05 1.78 3.59 0.00000596 0.00375 < 0.000001

10403.69 0.47 1.43 0.000225 0.000331 < 0.000001

8297.66 0.95 1.62 0.000225 0.176 < 0.000001

4570.52 4.96 7.49 0.000296 0.106 < 0.000001

9440.24 0.71 1.08 0.000296 0.11 < 0.000001

6656.68 6.27 2.07 0.000608 0.00000986 < 0.000001

11325.26 0.67 1.09 0.00168 0.488 < 0.000001

9172.09 34.55 12.13 0.00171 0.00017 < 0.000001

7042.18 23.45 5.42 0.00252 0.0000902 < 0.000001

6020.90 5.33 2.28 0.00252 0.00017 < 0.000001

11185.69 0.75 1.19 0.00252 0.891 < 0.000001

4995.04 2.28 3.87 0.00409 0.162 < 0.000001

8456.35 1.25 2.82 0.00517 0.31 < 0.000001

9960.78 0.78 1.82 0.0059 0.211 < 0.000001

7283.97 39.59 7.2 0.00715 0.000151 < 0.000001

7249.02 2.69 1.77 0.00757 0.0042 0.00064

7173.62 4.21 7.46 0.00935 0.647 < 0.000001

5711.21 4.23 6.21 0.00935 0.441 < 0.000001

9378.33 1.82 1 0.00935 0.00238 < 0.000001

4048.24 4.11 7.16 0.0141 0.176 < 0.000001

5837.80 2.19 3.04 0.0197 0.423 < 0.000001

3907.34 41.93 14.71 0.021 0.00771 < 0.000001

10840.50 0.99 2.24 0.0231 0.149 0

6745.30 2.08 1.56 0.0392 0.0221 0.00817

Waloszczyk et al. Diagnostic Pathology 2011, 6:22http://www.diagnosticpathology.org/content/6/1/22

Page 3 of 7

ResultsMS analysis in 100 (15 control, 85 pathological) tissueisolates was done. Results as combined spectra in gelview mode are shown in figure 1.All the spectra were analyzed with PCA algorithm and

resulting data distribution in three-dimensional spaceview is presented in figure 2. Rounded points are thecontrol group profiles and are marked for reasons ofclarity.Statistical analysis of all profiles for biomarker discovery

revealed 39 masses of statistical importance (p < 0.005)

which influenced division of investigated groups. Resultsare listed in table 1.Proteome profiles were classified using three different

supervised algorithms. Data for both (X1 - control, X2 -pathological) and recognition capability are listed intable 3.

DiscussionThe proteomic investigation as a novel technique forbiomarker discovery in disease process requires applica-tion of dedicated solutions including both advancedinstruments and software. However, separation and con-centration procedure are crucial for protein analysis[16]. Proteins of interest are a minority in comparisonto a huge number of large structural proteins and othercompounds which may negatively influence identifica-tion process especially in complex biological materials(blood, tissue) [17]. Moreover, mass spectroscopy withmatrix associated laser desorption ionization, has itslimitations: lack of highly abundant proteins in a sample,lack of salts, need for appropriate analytic concentra-tions, etc [18]. A number of protocols are described inliterature for separation, concentration and desalting ofproteins, however most of them are costly and timeconsuming [19-21]. We found that for peptides/proteinsprofiles purposes, considering the nature (composition -structure) of materials, it was enough to isolate only alimited range of proteins which could be presented as asingle point in multidimensional space basic on

Table 2 Types f pathological change diagnosed inexamined material

Diagnosis Number of cases

Squamous cell carcinoma 25

Adenocarcinoma 27

Large cell carcinoma 4

Typical carcinoid 1

Large cell neuroendocrine carcinoma 3

Small cell carcinoma 2

Adenosquamous carcinoma 4

Sarconiatoid carcinoma 4

Bening tumors 2

Metastasis 2

Lymphoma 2

Ectopic tissue and tumor-like lesions 9

Figure 1 MS spectra in gel view mode: a) control group, b) pathologically changed tissues.

Waloszczyk et al. Diagnostic Pathology 2011, 6:22http://www.diagnosticpathology.org/content/6/1/22

Page 4 of 7

statistical methods. A large number of mass signalsrecorded during measurement merge into one “signal”(a point located in multidimensional space) which canbe treated as a specific biomarker. The obtained profiles(single points) become characteristic “fingerprints” fortissue composition. Any pathological changes in theexamined material result in different localization incomparison to unchanged material what can be easyseen. Depending on the extent of pathological changesin tissue structure, proteins isolation procedure shouldinclude either more or fewer separation steps to obtainfractions which then can be successfully analyzed sepa-rately. In our approach we performed a simple isolationprocedure based on precipitation of large proteins and

then filtration using filters of cut-off level 3 kDa to con-centrate and desalt peptides/proteins. This procedureallowed us to collect proteins in such a form that theycan be measured directly in spectrometer without anyexpensive and time consuming procedures. As a detailexamination of structure of particular peptides/proteinsdifferentiating points distribution in multidimensionalspace was not our intent, we focused on reproducibilityof this procedure. Our aim was to analyse profiles andcompare them rather then to analyse a particular com-pound of which the profile was composed. Figure 1 pre-sents spectra of all 100 samples, including controltissues (A) and pathologically changed (B) in gel viewmode in order to combine all of them in one Figure 1.The data combination reveals that common signals maybe noted in broad spectrum for both groups but patho-logical samples are even more complex in rangesbetween common bands. Signals in the same masses aremost likely typical for the tissue origin (lung) and theyreflect common to some extend structure for both typesof the same tissues (pathological and normal). The mul-tivariate statistical method enabled to distinguish normalsamples from those diagnosed with cancer as well asthose which indicate other pathological changes (table2). As we were aware that signals typical for pathological

Figure 2 The result of principle component analysis (PCA). Data distribution in three-dimensional space (PC1 - PC2- PC3). Points of controlgroup were rounded.

Table 3 Classification results

Algorithm Validation Recognitioncapability

XVal X1 X2

Support Vector Machine(SVM)

95.8% 98.8% 92.9% 100%

Genetic Algorithm (GE) 95.3% 97.7% 92.9% 100%

Supervised Neural Network(SNN)

88.7% 98.8% 78.6% 100%

Waloszczyk et al. Diagnostic Pathology 2011, 6:22http://www.diagnosticpathology.org/content/6/1/22

Page 5 of 7

tissue may overlap with signals coming from normal tis-sue which is present in material in every tumour growthin an unchanged form, we decided to examine a rela-tively large tissue segment of 400 mg, covering normaland pathological region. This facilitated testing specifi-city and sensitivity of the method with view to an earlypathological change detection (in case the change can-not be seen macroscopically).Recorded MS data typically comprise a large number

of signals which makes it impossible to tell the differ-ence between examined groups basing on direct raw MSspectra. However, the principle component analysis(PCA) algorithm enables to present complex spectrumin a single point located in multidimensional space. Thislocation is strictly defined by sample composition, andthe points representing spectra can be easily compared.The result of PCA analysis of all the samples shown inFigure 2 revealed different localization of normal andpathological samples. Moreover, pathological sampleswere more dispersed in space which could be explainedby the fact that many different types of cancer wereexamined. We found 39 different masses (table 1) whichare of statistical importance and discriminate all thedata. The examination of compounds structure is a typi-cal approach and has been widely described in refer-ences. We found that as much as 39 masses influencesamples distribution in multidimensional space and allof them may be important for change identification.A typical approach based on a single, one-parametricalstatistical investigation seems to be less efficient com-pared to multiparametrical methods especially in respectof sensitivity and selectivity. Also we attempted to pointout that there was no need to precisely define (proteinstructure identification) each of 39 signals because all ofthem are strictly defined by the point located in multidi-mensional space (non - dimensional data x, y, z). Pointsobtained from PCA calculations can be then classifiedand compared using different classification algorithms(table 3). In our experiment the recognition capability ofthe tree most common models was 100% which maydemonstrate a high specificity of the procedure.

ConclusionsThe proteomic investigation of tissue samples usingadvanced statistical methods enabled differentiation oflung tissue with cancer and unchanged tissue materialeven in an early stage of growth (low pathologcial tounchanged tissue ratio in sample). Following ist valida-tion, this method, supporting commonly used histologi-cal techniques, may be considered an objective, preciseand fast procedure in an early cancer diagnosis. Ourprotocol may be useful in procedures that need a largenumber of samples due to the elimination of time con-suming steps. Multiparameter statistical methods may

be helpful in elaboration of newly defined biomarkersfor clinical applications expressed as many factors “inone point”.

Author details1Department of Toxicology and Molecular Pathobiochemistry, PomeranianMedical University, 70-204 Szczecin, Poland. 2Professor A. SokolowskiSpecialist Hospital Szczecin Zdunowo, Poland.

Authors’ contributionsPW collected and registered all the samples, performed histologicalexaminations, participated in the design of the study, MS measurements anddata analysis, participated in data analysis and interpretation. TJ conceived ofthe study, developed the methodology, carried out MS measurements,performed the data analysis and interpretation, drafted the manuscript. JAand TG collected segments of specimens and participated in clinicaldiagnosis. KB participated in design of the study and its coordination. Allauthors read and approved the final manuscript.

Competing interestsThe authors declare that they have no competing interests.

Received: 27 November 2010 Accepted: 21 March 2011Published: 21 March 2011

References1. Palmblad M, Tiss A, Cramer R: Mass spectrometry in clinical proteomics -

from the present to the future. Proteomics Clin Appl 2009, 3:6-17.2. Munro NP, Cairns DA, Clarke P, Rogers M, Stanley AJ, Barrett JH,

Harnden P, Thompson D, Eardley I, Banks RE, Knowles MA: Urinarybiomarker profiling in transitional cell carcinoma. Int J Cancer 2006,119:2642-2650.

3. Coombes KR, Morris JS, Hu J, Edmonson SR, Baggerly KA: Serumproteomics profiling - a young technology begins to mature. NatureBiotechnol 2005, 23:291-292.

4. Check E: Proteomics and cancer: running before we can walk. Nature2004, 429:496-497.

5. Zhang X, Leung SM, Morris CR, Shigenaga MK: Evaluation of a novel,integrated approach using functionalized magnetic beads, bench-topMALDI-TOF-MS with prestructured sample supports, and patternrecognition software for profiling potential biomarkers in humanplasma. J Biomol Technol 2004, 15:167-175.

6. Li J, Zhang Z, Rosenzweig J, Wang YY, Chan DW: Proteomics andbioinformatics approaches for identification of serum biomarkers todetect breast cancer. Clin Chem 2002, 48:1296-1304.

7. Petricoin EF, Ardekani AM, Hitt BA, Levine PJ, Fusaro VA, Steinberg SM,Mills GB, Simone C, Fishman DA, Kohn EC, Liotta LA: Use of proteomicpatterns in serum to identify ovarian cancer. Lancet 2002, 359:572.

8. Conrads TP, Zhou M, Petricoin EF, Liotta LA, Veenstra TD: Cancer diagnosisusing proteomic patterns. Expert Rev Mol Diagn 2003, 3:411.

9. Aebersold R, Goodlett DR: Mass spectrometry in proteomics. Chem Rev2001, 101:269-295.

10. Stoeckli M, Staab D, Schweitzer A: Compound and metabolite distributionmeasured by MALDI mass spectrometric imaging in whole-body tissuesections. Int J Mass Spectrom 2007, 260:195-202.

11. Andersson M, Groseclose MR, Deutch AY, Caprioli RM: Imaging massspectrometry of proteins and peptides: 3 D volume reconstruction. NatMethods 2008, 5:101-108.

12. Caldwell RL, Caprioli RM: Tissue profiling by mass spectrometry: A reviewof methodology and applications. Mol Cell Proteomics 2005, 4:394-401.

13. Chaurand P, Norris JL, Cornett DS, Mobley JA, Caprioli RM: Newdevelopments in profiling and imaging of proteins from tissue sectionsby MALDI mass spectrometry. J Proteome Res 2006, 5:2889-2900.

14. Cornett DS, Reyzer ML, Chaurand P, Caprioli RM: MALDI imaging massspectrometry: Molecular snapshots of biochemical systems. Na Methods2007, 4:828-833.

15. Khatib-Shahidi S, Andersson M, Herman JL, Gillespie TA, Caprioli RM: Directmolecular analysis of whole-body animal tissue sections by imagingMALDI mass spectrometry. Anal Chem 2006, 78:6448-6456.

Waloszczyk et al. Diagnostic Pathology 2011, 6:22http://www.diagnosticpathology.org/content/6/1/22

Page 6 of 7

http://www.ncbi.nlm.nih.gov/pubmed/21136932?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/21136932?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/16991122?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/16991122?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/15175721?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/12142387?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/12142387?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/12142387?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/11867112?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/11867112?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/12877381?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/12877381?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/11712248?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/18165806?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/18165806?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/15677390?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/15677390?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/17081040?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/17081040?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/17081040?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/16970320?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/16970320?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/16970320?dopt=Abstract

16. Rifai N, Gillette MA, Carr SA: Protein biomarker discovery and validation:The long and uncertain path to clinical utility. Nat Biotechnol 2006,24:971-983.

17. Duncan R, McConkey EH: How many proteins are there in a typicalmammalian cell? Clin Chem 1982, 28:749-755.

18. Diamandis EP: Mass spectrometry as a diagnostic and a cancerbiomarker discovery tool: opportunities and potential limitations. MolCell Proteomics 2004, 3:367-378.

19. Omenn GS: Strategies for plasma proteomic profiling of cancers.Proteomics 2006, 6:5662-5673.

20. Rifai N, Gillette MA, Carr SA: Protein biomarker discovery and validation:The long and uncertain path to clinical utility. Nat Biotechnol 2006,24:971-983.

21. Görg A, Weiss W, Dunn MJ: Current two-dimensional electrophoresistechnology for proteomics. Proteomics 2004, 4:3665-3685.

doi:10.1186/1746-1596-6-22Cite this article as: Waloszczyk et al.: Proteomic patterns analysis withmultivariate calculations as a promising tool for prompt differentiationof early stage lung tissue with cancer and unchanged tissue material.Diagnostic Pathology 2011 6:22.

Submit your next manuscript to BioMed Centraland take full advantage of:

• Convenient online submission

• Thorough peer review

• No space constraints or color figure charges

• Immediate publication on acceptance

• Inclusion in PubMed, CAS, Scopus and Google Scholar

• Research which is freely available for redistribution

Submit your manuscript at www.biomedcentral.com/submit

Waloszczyk et al. Diagnostic Pathology 2011, 6:22http://www.diagnosticpathology.org/content/6/1/22

Page 7 of 7

http://www.ncbi.nlm.nih.gov/pubmed/16900146?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/16900146?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/7074868?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/7074868?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/14990683?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/14990683?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/16991194?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/16900146?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/16900146?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/15543535?dopt=Abstracthttp://www.ncbi.nlm.nih.gov/pubmed/15543535?dopt=Abstract

AbstractBackgroundMethodsResultsConclusion

BackgroundMethodsMaterialsMethods

ResultsDiscussionConclusionsAuthor detailsAuthors' contributionsCompeting interestsReferences